I suppose the vast majority of training data used for cutting edge models was created after 1900.

Ofc they are because their primary goal is to be useful and to be useful they need to always be relevant.

But considering that Special Relativity was published in 1905 which means all its building blocks were already floating in the ether by 1900 it would be a very interesting experiment to train something on Claude/Gemini scale and then say give in the field equations and ask it to build a theory around them.

His point is that we can't train a Gemini 3/Claude 4.5 etc model because we don't have the data to match the training scale of those models. There aren't trillions of tokens of digitized pre-1900s text.

How can you train a Claude/Gemini scale model if you’re limited to <10% of the training data?

I don't know if this is related to the topic, but GPT5 can convert an 1880 Ottoman archival photograph to English, and without any loss of quality.

My friend works in that period of Ottoman archives. Do you have a source or something I can share?

and easier to blame parenting than a society where those with mental illness issues are left on their own.

You theoretically could blame bad parenting if the parent monopolized the time of the child to near 100% of their life. But, that isn't the case in our world today. Society, and the circle of people around, are most of the influence that shape the child's sense of reality.

The parent doesn’t have to get the kid a phone you know

The article ends on a note which can be read in either a hopeful or ominous key.

"Authorities anticipate no significant conflicts with the beaver colony for at least the next decade."

2036: Czech authorities sue for peace after months of devastating conflict against the beaver colony.

We have always been at war with Eurasian beavers (Castor fiber).

Farmers have been at war with them. Local governments usually side with the beavers, because their longterm benefits are higher, but in right-wing crisis times the farmers will win.

I'm not sure I trust the people who get more work to do, more budget and more power out of the beavers being there anymore than I trust the farmers who would be better off with the beavers gone.

A painter will always tell you you need paint.

A roofer will always try and sell you a roof.

A tire shop will always try and sell you tires.

But we can take these people at face value?

If you want peace, prepare for war.

I wonder if text written using chinese characters is more compatible with such vision centric compression than latin text.

I think it's not the case. Chinese characters have the highest information entropy of all writing systems. However, Chinese characters are all independent symbols, which means if you want the LLM to support 5000 Chinese characters, you need to put 5000 characters into the lookup table (obviously there's no root, prefix, and suffix in Chinese, you cannot split the character into multiple reusable word pieces). As a result, you may need fewer characters to represent the same meaning compared to latin languages, but LLMs may also need to activate more token embeddings.

"Helix doesn’t have tabs, instead it has a nice buffer switcher (<space>b) I can use to switch to the buffer I want"

It has tabs if you add (the completely undiscoverably named) bufferline="multiple" to the [editor] section in config.

Does HTMX have a pro version? Is it dead?

Far from dead. Usage is growing.

HTMX is a single htmx.js file with like 4000 lines of pretty clearly written code.

It purports to - and I think succeeds - in adding a couple of missing hypermedia features to HTML.

It's not a "framework" - good

It's not serverside - good

Need to add a feature? Just edit htmx.js

Some people will get a fit once they find out it's JS, and not TypeScript.

No body is preventing them to write a layer of ts definition on top of the js. Check out postgres.js

It will be, eventually, unless a maintainer is able to maintain during the day. It doesn't matter what the source of free time is however: retired, rich, runs a company from their open source project, paid by somebody else, etc., but full time job + open source maintainer = dead project, eventually.

htmx isn't dead, but, from a semantic stand point, it's pretty done:

https://htmx.org/essays/future/

there are bugs, but we have to trade off fixes against potentially breaking users sites that depend on (sometimes implicitly) the current behavior

this makes me very hesitant to make changes and accept PRs, but i also feel bad closing old issues without really diving deep into them

such is life in open source

I don't think open issues is a fair way to judge project liveness. TypeScript also has hundreds of open issues going back years with no traction. Is TypeScript dead?

Yes, issues that are years old show me the commitment level. Not a knock against HTMX but a clear sign of priorities. Carson is free to meme all day and talk about other projects. It's very clear where he stands and that's fine

this year I created and released fixi.js, created the montana mini computer (https://mtmc.cs.montana.edu), published an paper on hypermedia via the ACM, got hyperscript to 1.0, released 3 versions of htmx, reworked all the classes that I teach at montana state and am planning on releasing a java-based take on rails that I'm building for my web programming class

i am also the president of the local youth baseball program and helped get BigSkyDevCon over the hump

i think you'd be surprised at how little time i actually spend on twitter

as always, my issue is never with how you spend your time. you are a giver of gifts and I wish more people that relied on HTMX stepped up to make it better. in no way should anything be expected of you. How you spend your time is obviously your call. MIT is MIT

It was a rhetorical question; the answer is no, old issues with no updates don't necessarily indicate anything about the health of the project. Different people have different project management styles. You use your style for your project, and Carson uses his for htmx. There's no one correct way to manage an issue backlog.

My project is really healthy then, as I summarily close issues as “not planned” rather than leave them open.

> I don't want my products looking anything like other websites.

if your product is a personal page or art/game, sure, understandable. Otherwise for apps it is beneficial to have consistent UI. It was the case for decades on desktops, it is true to a certain degree on mobile phones, and it makes users' lives easier.

I remember fighting this one well over a decade ago when management was telling us engineers that our web, Android and iOS should all look, feel and behave the same; it took some time to convince them that what you need is not consistency across platforms but consistency _within_ platforms.

Nobody on iOS cares what/how your app looks/works on Android, they care that the UX meets the expectations they have of that OS because they're switching between apps on that platform all day every day, people actually moving Android<-->iOS are few and far between, I mean literal decade(s) time-frames that people aren't switching.

You don't have multiple devices? I have an Android phone, an iPad, a PC, a Mac, etc.

I have Windows, Linux with various desktops, iPhone, Android, and of course web browsers, and I think the apps look and behave pretty much the same across all devises. Maybe because almost all apps are web apps in a native shell. UI components it seems was a matter of performance rather then usability and developer experience!? Or it's just scripting madness gone framework insane.

Yes and they behave differently which is what I expect - they are different tools and they should behave consistently on a platform.

They should be tuned for the platform.

I can't use gestures on a PC or Mac but I can on a iPad or Android. Similarly I can control a PC or mac from a proper keybord and mouse but the usual use for iPad/Android is via a single finger.

I think you don’t find many billionaires with opposite ideology.

That's what makes not doing it moral